On Saturday 13 July 2002 21:25 pm, civileme wrote: > poogle wrote: > >The scenario, my Partner's standalone PC at work regularly crashing in Win > > 95, the solution I suggested was that I would install Mandrake, solution > > accepted and project complete. > >Day 1 - complete freeze while left unattended, no response to any key or > >mouse, clock showing the wrong time (the time it froze). > >Day 2 Having had nagging doubts about their power supply She is set up > > with my UPS. > >Day 3 & 4 No problems - Power supply then > >Day 5 Freeze > >Day 6 (today) PC to my workshop (well the spareroom really), run memtest > > for 9 cycles and only get one error during run 1 this is :- > >FAILURE: 0x00000000 != 0xffffffff at offset 0x005c5328 > >All other 8 cycles produced no errors. > >The question is does this indicate failed/failing memory ? > >I have never used memtest before so am unsure what I should expect (also > >unsure whether 9 cycles was enough or should I have left it running until > > it got bored and quit) > > > > > >------------------------------------------------------------------------ > > > >Want to buy your Pack or Services from MandrakeSoft? > >Go to http://www.mandrakestore.com > > One failure is one too many. > > REmove the memory, burnish contacts with eraser end of pencil, degrease > with isopropyl alcohol, clear slots with compressed air can, reinsert. > Run memory test again. > > Or alternatively, replace memory and run memory test again. It may be > the motherboard circuitry. > > But a freeze like that, with all the power management stuff turned off, > is indicative of a hardware problem. If the BIOS has ACPI and it is > activated, that may be the cause of a freeze when left unattended for a > while. > > But yes, the extremely low probability failure is unacceptable in > computing. If something has a finite chance of failure, no matter how > small, that allows you to calculate how often to expect it. There is no > doubt whatsoever that it will occur. There is no "almost" passing a > memory test. > > Windows is beloved of hardware manufacturers for that very reason. > Freeze-ups are so common from software causes that hardware flaws are > masked. Not so with linux. A freeze unrecoverable is a rare > occurrence. Even "frozen" boxes seem to be able to switch to console or > to allow an ssh entry or the emergency alt-sysrq-r alt-sysrq-s > alt-sysrq-b to go to raw keyboard input, emergency disk sync and reboot > without a reset switch. > > But here is what I have (last reboot was to change hardware) > > [tester@v5 tester]$ uptime > 12:16pm up 16 days, 21:14, 3 users, load average: 0.00, 0.01, 0.00 > [tester@v5 tester]$ > > [tester@msn75 tester]$ uptime > 12:08pm up 45 days, 19:41, 2 users, load average: 0.49, 0.17, 0.06 > [tester@msn75 tester]$ > > These are two of my systems, the second an IBM PC300GL P2-300, the other a > Duron 900 on a POJ Matsonic motherboard with the most buggy VIA chipset > ever. > > The gateway I could post as well, but that was last altered before the > release of 8.2 and is running with the latest updates, so it isn't fair, > but its uptime is more than 180 days, because I last booted it December 24, > 2001 (yep on my birthday when I changed its physical location and the > location of the DSL it was connected to. I did not even boot it when I > stopped using DSL and switched to cable modem. It is another IBM PC. > > Yes, all of these have passed memtest left for 3-4 days running, since they > are production machines. The last downtime I had was on the Matsonic Mobo > because a plastic shaving jammed the CPU cooling fan. The fan was replaced > and we are still running. > > Civileme > Thanks, that seems to have done the trick, 4 runs now and no errors. (The contractor comes tomorrow to take away the dust I blew out of it)
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com